An Integrated Architecture for Fault Diagnosis and Failure Prognosis with an Application to Aircraft Systems (slides)
Abstract This presentation is introducing an integrated fault diagnosis and failure prognosis methodology as applied on-line to flight critical aircraft systems. Enabling technologies include vibration and flight regime data pre-processing, de-noising algorithms to improve signal to noise ratio, feature or condition indicator selection and extraction routines, and model-based fault diagnosis and failure prognosis algorithms for accurate detection and robust prediction of the remaining useful life of failing components. We elaborate on the design of degradation or fatigue models using finite element analysis and nonlinear dynamic models in the frequency domain in order to gain a better understanding of the physics of failure mechanisms, to extract an optimum feature vector and to establish a basis for incipient failure detection and prediction. The modeling framework is coupled with Bayesian estimation techniques, specifically particle filtering, for accurate diagnosis and prognosis. Seeded fault data from a testcell and fleet aircraft data are used to train and validate the algorithms. A Graphical User Interface and computer code for all processing algorithms are implemented on conventional health monitoring apparatus available on several military aircraft. Results from "blind" testing demonstrate the efficacy of the architecture.
Bio Professor Vachtsevanos was born in Kozani, Greece. He attended the City College of New York and received his B.E.E. degree in 1962. He received an M.E.E. degree from New York University and his Ph.D. degree in Electrical Engineering from the City University of New York in 1970. His research focused on adaptive control systems. After graduate school, he taught within the City University System of New York from 1970 through 1975; from 1975 through 1977 he served as an Associate Professor of Electrical Engineering at Manhattan College. In 1977 he was elected to a chair professorship in Electrical Engineering at the Democritus University of Thrace, Greece, where he served as the Pro-Rector during the 1978-79 academic year. He joined the Georgia Institute of Technology as a Visiting Professor in 1982-1983 and as a Professor in 1984. He is also serving as an adjunct faculty member in the School of Textile and Fiber Engineering. He was the 3M McKnight Distinguished Visiting Professor at the University of Minnesota, Duluth during the 1994 Spring Semester. Since joining the faculty at Georgia Tech, Dr. Vachtsevanos has been teaching courses and conducting research on intelligent systems, robotics and automation of industrial processes and diagnostics/prognostics of large-scale complex systems.
Health Management Technology Integration (slides)
Abstract Academia, industry and government research laboratories have made great strides in developing and maturing software and hardware technology to monitor and even predict the health of machines. The output of health monitoring (i.e., diagnosis, prognosis, usage) are further processed and integrated with machine controls, operations, maintenance and logistics to produce a health management capability with the end goal of enabling business to better utilize their assets. This presentation examines some of the challenges or hurdles to the integration of health monitoring/management technology into aircraft fleet operations. Key focus areas or points of integration include creating the business case, consideration of health management in the design process, integrating health monitoring hardware and software into the aircraft, leveraging the resulting output into the operations and support decision loop, handling the additional data, addressing regulatory concerns, and the allocation of the responsibilities and roles of the parties involved in the development, use and maturation of a health management system. Boeing Phantom Works Support and Services thrust is tasked with facilitating the maturing health management technology for integration into Boeing products. The presentation also describes some of the efforts by the Phantom Works to address the health management integration challenges above including the creation of a Health Management Engineering Environment. This environment consists of three laboratories: Program Analysis and Modeling to establish the business case, Development Laboratory to provide the tools and processes, and the Operations Laboratory to perform integrated demonstrations in realistic environments.
Bio Dr. Kirby Keller is a Technical Fellow for the Boeing Company. He has 33 years experience in the development of mission and vehicle health management systems to air, ground and space vehicles. His current duties include Technical Lead for the Integrated Vehicle Health Management programs within the Support and Services Thrust of Boeing Phantom Works. In this position, he is the principal investigator and system designer for several internal research and development projects that are focused on developing a Boeing wide common architecture and best practices for IVHM. He is program manager for the USAF sponsored Dual Use Science and Technology (DUST) Aircraft Electrical Power System Prognostics and Health Management (AEPHM) program. He was the technical lead for the Navy Open System Architecture for Condition Based Monitoring research program and for the USAF/Boeing Unmanned Combat Air Vehicle (UCAV) System prognostics and health management risk reduction demonstrations. Other, past programs activity includes the On-line Flight Control Diagnostic System sponsored by the Navy and U. S. Army Rotorcraft Pilot's Associate program. He holds a PhD in Mathematics from Iowa State University and is the author of over 40 technical papers, conference presentations, and technical reports.